Cancers contain altered methylation patterns that result in aberrant expression of critical genes. Hypermethylation turns off expression of genes required to regulate normal growth while hypomethylation allows for inappropriate expression of genes that allow cells to proliferate. Promoters for genes often have regions of high CpG content known as “CpG Islands”. When genes, such as tumor suppressor genes, with promoter CpG islands are turned off, this is usually accompanied with methylation of most CpG sequences within the promoter and 1st intron regions. Aberrant promoter hypermethylation occurs at the 5-position of cytosine within the CpG dinucleotide (Gardiner-Garden et al., J. Mol. Biol., 196(2): 261-82 (1987)). It inactivates the expression of critical genes that are involved in tumor suppression, DNA repair, control of tumor metastasis, and invasion (Cheng et al., Genome Res. 16(2): 282-89 (2005), Feinberg et al., Nature, 301: 89-92 (1983); Jones et al., Nat. Rev. Genet., 3(6): 415-28 (2002)). There is a great need in both basic and clinical research to identify promoter DNA methylation status with high efficiency and accuracy for disease diagnoses and prognoses.
The presence and absence of methylation in certain genetic regions has prenatal diagnostic and prognostic applications. For example, aberrant methylation on regions on chromosomes 13, 18, 21, X, and Y can be used to diagnose Down syndrome (Patsalis et al., Exp. Opin. Biol. Ther. 12 (Suppl. 1): S155-S161 (2012). Because fetal DNA and maternal DNA are differentially methylated, cell-free DNA in maternal plasma can provide a source of fetal DNA, which can be obtained non-invasively and utilized to assess the methylation state of the aforementioned chromosomes.
Currently, a number of groups have used bisulfite approaches to detect the presence of low levels of methylated DNA in serum, as a marker of early cancer (deVos, Clinical Chemistry 55(7):1337-1346 (2009), Lind et al., Molecular Cancer 10:85 (2011)). However, often a single marker gives unacceptably high false-positive and false-negative results (Alquist et al., Clin. Gastroenterol. Hepatol. 10(3): 272-277 (2012)). Thus, a single or a few methylation markers is insufficient for robust detection of early cancer from the serum. There is an urgent need for methods with multiplexed detection of very low levels of methylated DNA when the majority of DNA with the same sequence is unmethylated. For example, detection of multiple methylated DNA sequences in cell-free DNA isolated from serum may enable early detection of cancer. Likewise, methods for multiplexed detection of very low levels of unmethylated DNA when the majority of DNA with the same sequence is methylated are also urgently needed for applications such as early detection of cancer.
Various methods have been developed for the study of promoter DNA methylation status of known genes (Laird P. W., Nature Review Cancer, 3: 253-266 (2003)). These methods can generally be grouped into two categories: methylation-sensitive restriction endonuclease assays and sodium bisulfite conversion based approaches.
Methylation-Sensitive Restriction Endonuclease Digestion Methods:
This approach takes advantage of methyl-sensitive restriction enzymes, wherein genomic DNA is cleaved when unmethylated, and this is followed by a PCR amplification using primers that flank the site(s) (Singer-Sam et al., Nucleic Acids Res., 18(3): 687 (1990), Singer-Sam et al., Mol. Cell. Biol., 10(9): 4987-9 (1990)). A methylated restriction endonuclease site results in the presence of the proper PCR product. The credibility of this method depends on the complete digestion of unmethylated DNA by the restriction endonuclease. This problem is exacerbated by: (i) limiting amounts of methylated DNA in the sample, (ii) the requirement of some restriction enzymes to bind two unmethylated sites simultaneously, and (iii) the lack of, or poor activity of restriction enzymes to single-stranded DNA that may arise during sample preparation. It is difficult to drive endonuclease digestions to completion. Thus, it is sometimes difficult to determine whether PCR amplicons result from incomplete digestion (i.e. false positives) or from those of low abundance methylation sites (i.e. true positives). Restriction enzyme techniques are based on removing the unmethylated DNA, and assuming that PCR amplification of the remaining DNA arises because it was methylated, and consequently the method is susceptible to false positives arising from incomplete removal of unmethylated DNA. This technique has the disadvantage that it is not accurate for finding low levels of methylated DNA when the majority of the same sequence is unmethylated, as would be the case with detection of cancer-associated methylation at multiple markers in cell free DNA from the serum.
Sodium-Bisulfite-Based Chemical Conversion.
Chemical conversion of cytosines to uracils using bisulfite can be used to detect DNA methylation differences. 5-methylcytosines are resistant to conversion, and deamination only occurs on unmethylated cytosines (Frommer et al., Proc. Natl. Acad. Sci. USA, 89(5): 1827-31 (1992)). Bisulfite can be quantitatively added to the 5-6 double bonds of cytosine if there is no methyl group on the 5 position. Bisulfite addition renders the cytosine susceptible to hydrolytic deamination; subsequent elimination of the bisulfite results in the formation of uracil (Voss et al., Anal. Chem., 70(18): 3818-3823 (1998)). One strand of the modified DNA sequences can then be PCR amplified and sequenced. However, due to stromal cell contamination in a typical clinical sample, direct sequencing without cloning the PCR products reduces the sensitivity of the technique. It requires about 25% of the alleles to be methylated for accurate detection (Myohanen et al., DNA Sequence, 5: 1-8 (1994).
Development of methylation-specific PCR (MSP) has allowed the sensitive and specific study of low abundance methylation sequences (Herman et al., Proc. Natl. Acad. Sci. USA, 93(18): 9821-6 (1996)). MSP relies upon chemical modification of DNA using bisulfite, and specifically designed PCR primers that are complementary to the bisulfite modified DNA template. Typically, more than three CpG sites have to be included in the oligonucleotide sequences. Two sets of MSP PCR primers are designed, one set of the MSP primers has the sequence to perfectly hybridize to the complementary strand of the bisulfite-treated methylated DNA sequence with methyl-cytosines residing on the CpG sites. The other set of the MSP primers is only designed to perfectly hybridize to the complementary strand of the bisulfite-treated DNA sequence in the absence of methylated cytosine. Consequently, the MSP specific PCR products only results from the DNA template which contains methyl-cytosines.
There are three major difficulties with this approach. The design of MSP primers requires sufficient numbers of methylated cytosines to be present in the primer sequence to ensure the selection capability. It may not be sufficiently sensitive to distinguish partial methylated sequences from fully methylated one. In addition, this assay analyzes one gene at a time, and both sets of MSP primers have different annealing temperatures which may further slowdown its throughput. Finally, bisulfite treatment of DNA often nicks the DNA (i.e. destroys the backbone chain) as it is also converting unmethylated cytosines to uracil. Conditions which assure that all unmethylated cytosines are converted to uracil may also destroy the DNA. Conditions which assure that sufficient DNA remains intact may not assure that all unmethylated cytosines are converted to uracil. Thus, absence of a band may be the consequence of destroying too much of the starting DNA and, consequently, insufficient amplification, leading to a false negative result. Likewise, presence of a band may be the consequence of incomplete conversion of unmethylated cytosine to uracil, allowing for primer binding at an unmethylated site, and leading to a false positive result. Some of these problems may be overcome by combining the use of Bisulfite treatment, the polymerase chain reaction, and the ligase detection reaction (see U.S. Pat. No. 7,358,048 to Barany et al.)
A further improvement of this technique employs a blocking oligonucleotide that hybridizes to the sequence for bisulfite-converted unmethylated DNA, thus enriching for amplification of bisulfite-converted methylated DNA (deVos et al., Clinical Chemistry 55(7):1337-1346 (2009)). The disadvantage is that bisulfite treatment destroys from 50% to 90% of the original DNA integrity by nicking it. When starting with DNA from the serum (with average length of about 160 bases), this can be a significant problem. Further, converting C's to U's reduces the complexity of the sequence from 4 bases to 3 bases. Thus, non-specific amplifications can occur. This usually necessitates a nested-PCR approach; this runs the risk of carryover contamination and is generally not ideal for multiplexed amplifications.
The present invention is directed at overcoming this and other deficiencies in the art.